This article is mentioned in:

Introduction

Female breast cancer is a common malignancy globally (1). Triple-negative breast cancer (TNBC), making up 12–17% of cancers of the breast, is defined by the absence of receptors for oestrogen, progesterone, and human epidermal growth factor receptor 2 (HER2) on the tumour cells (2–4). TNBC is the most aggressive form of breast cancer, and rates of recurrence, distant metastasis, and mortality are significantly higher than for other types of breast cancer (2,3). Part of the reason for this may be a more limited range of treatment options for TNBC than for these other types of breast cancer.

Over the last decade, cancer immunotherapy has become established as a highly effective treatment modality for certain cancers (5). Immune checkpoint blockade (ICB) has achieved remarkable clinical results in some patients with malignant melanoma, renal cancer, non-small cell lung cancer, and other solid tumours (6). Currently, treatment with antibodies to programmed death-ligand 1 (PD-L1) or programmed death-1 (PD-1) is the mainstay of ICB, and has also been investigated for TNBC, with some degree of success. Thus, the IMpassion 130 trial (NCT02425891) used anti-PD-L1 (atezolizumab) together with nab-paclitaxel as first-line treatment for advanced or metastatic TNBC and reported that this combination was superior to nab-paclitaxel alone (7). Additionally, KEYNOTE-355 investigated the efficacy of anti-PD-1 (pembrolizumab) combined with chemotherapy (nab-paclitaxel; paclitaxel; or gemcitabine plus carboplatin) and reported increased progression-free survival of TNBC patients relative to chemotherapy alone (8). Nonetheless, only 20–58% of TNBC tumours express PD-L1, making it likely that many TNBC patients will not experience any clinical benefit from ICB directed to this molecule (9–15). Also, repetitive administration can result in resistance to ICB, as with chemotherapy (16). Hence, there is an urgent unmet medical need for novel treatment targets in TNBC.

Potential ICB targets other than PD-L1 and PD-1 may also be considered for application in TNBC. One of these, immunoglobulin superfamily member, T-cell immunoglobulin mucin-3 (TIM-3), is a checkpoint molecule present on many different immune cells, including dendritic cells, macrophages, and T cells (17–20). TIM-3 mediates suppressive activity after binding a variety of different ligands, including phosphatidylserine, CEACAM-1, and galectin-9 (17,21,22). The latter is one of the family of h-galactoside-binding proteins which is over-expressed by many tumours; its binding to TIM-3 on T cell surface results in cytotoxic T cell suppression via an autocrine pathway (23–29). It has therefore been hypothesized that either or both TIM-3 and galectin-9 could represent novel therapeutic targets (30–32). However, the prognostic significance of these two molecules has not been unequivocally established, because their high expression has been reported to associate with either a better or worse prognosis, depending on the specific tumour entity (33–38). In the case of TNBC, TIM-3 or galectin-9 expression has been associated with certain clinicopathological features and with prognostic significance (34,38) but to the best of our knowledge, no studies to date have examined the relationship between TIM-3 in combination with one of its ligands, galectin-9. Therefore, we explored correlations between TIM-3 and galectin-9 expression in TNBC by immunohistochemistry, and investigated their impact on patient prognosis and clinicopathological features.

Materials and methods

Patient selection

Patients with TNBC undergoing surgical resection at the Department of Surgery of the Kansai Medical University Hospital between January 2006 and December 2018 were enrolled. Patients receiving neoadjuvant chemotherapy, known to influence TIM-3 and galectin-9 expression, were excluded. Inclusion criteria were invasive breast carcinoma of no special type according to the recent World Health Organization Classification of Breast Tumors (39), but those with special types were excluded because each of these has different clinicopathological features. Finally, 62 TNBC patients were included. This cohort is essentially identical to that described in our previous studies (40–43). To date, in this cohort, we have analysed associations between adipophilin expression and prognosis (40), as well as the prognostic impact of PD-L1 expression by cancer-associated fibroblasts (41), and relationships between PD-L1 and the expression of the immune checkpoint protein CD155 (42). We have also compared three different PD-L1 assays in patients with TNBC using immunohistochemistry (43). The focus of the current study was to determine the prognostic significance of TIM3 and galectin-9 expression in this same cohort of TNBC patients.

This is a retrospective single-centre study conducted in accordance with the principles of the Declaration of Helsinki. The study protocol was approved by the Institutional Review Board of Kansai Medical University Hospital (Approval #2019041). Because of the retrospective study design, informed consent was obtained using the opt-out method, there being no risk to the participants. Information on the study, including the inclusion criteria and the opportunity to opt-out, was made available on the institutional website (https://www.kmu.ac.jp/hirakata/hospital/2671t800000136cd-att/a1565060399005.pdf).

Histopathology

All histopathological diagnoses were independently evaluated by at least two experienced diagnostic pathologists, using the TNM Classification of Malignant Tumors, 8th Ed. Grading followed the Nottingham scale (44). Dichotomization of the Ki-67 labelling index (LI) was set as high at ≥40% and low at <40%, following a meta-analysis of patients with TNBC (45). Stromal tumour-infiltrating lymphocytes (TILs) were identified using haematoxylin and eosin staining and were considered lymphocyte-predominant breast cancer (LPBC) at ≥60% and non-LPBC at <59%, according to TIL Working Group guidelines (46,47).

Tissue microarrays

Regions most morphologically representative of carcinoma were selected by H&E staining of the slides, and for every patient, three 2 mm-diameter tissue cores were punched out of the paraffin-embedded blocks.

Immunohistochemistry

Immunohistochemistry used the Discovery ULTRA System (Roche Diagnostics, Basel, Switzerland) according to the manufacturer's instructions. Antibodies were as follows: TIM-3 (rabbit monoclonal antibody, D5D5R, Cell Signaling Technology, Danvers, MA, USA; diluted 1:200); galectin-9 (mouse monoclonal antibody, ab153673, 1G3, Abcam plc, Cambridge, UK; diluted 1:200). At least of two researchers independently evaluated the immunohistochemistry results. TIM-3-positivity was defined as membrane staining of any intensity on ≥1% of TILs (36). Galectin-9-positivity was defined as membrane staining of any intensity on ≥1% of tumour cells (48). The patient was classified as having TIM-3- or galectin-9-positive tumour when one or more cores from the same individual were positive according to this definition.

Statistical analysis

We used SPSS Statistics 27.0 (IBM, Armonk, NY, USA) for all analyses. Correlations between two groups were calculated using Fisher's exact test for categorical variables and Mann-Whitney U testing for continuous variables. Relapse-free survival (RFS) was determined using the Kaplan-Meier method, with log-rank testing. Cox proportional hazards modelling was used to estimate relationships between clinicopathological parameters and survival. Statistical significance was set at P<0.05.

Results

Patients

The cohort of 62 women with TNBC studied here is the same as described earlier (43). Their clinicopathological characteristics were presented in the previous publication (43). Briefly, median age at initial diagnosis was 68 years (range, 31–93); the diagnosis of TNBC relied on biopsy results. Patients with invasive carcinoma of no special type were selected (see Materials and Methods). Median tumour diameter was 21 mm (range, 2–55 mm). Median follow-up was 58 months (range, 11–173 months). Eleven (17.7%) patients relapsed, all with distant metastases. There were no local recurrences. Nine patients (14.5%) died of their disease.

Correlations between galectin-9 or TIM-3 expression and clinicopathological factors

Of the 62 patients, 49 (79.0%) were classified as galectin-9-positive (Fig. 1). Table I presents associations between galectin-9-positivity and clinicopathological factors. The use of adjuvant chemotherapy was associated with galectin-9 expression (P=0.040), but not with any other factors, including age and menopausal status. There were also no associations between galectin-9-positivity and the clinicopathological factors staging, Nottingham histological grade, lymph node status, lymphovascular invasion, Ki-67 LI, or stromal TILs.

Figure 1. Immunohistochemical staining for galectin-9. Positive staining is seen for the neoplastic cells (magnification, ×400).

Table I. Association between clinicopathological factors and galectin-9 expression.

Table I.

Association between clinicopathological factors and galectin-9 expression.

Factors	Galectin-9-positive (n=49)	Galectin-9-negative (n=13)	P-value
Median age ± SD, years	64±15	72±13	0.115
Menopausal status, n
Premenopausal	9	0	0.184
Postmenopausal	39	13
Unknown	1	0
Tumour size, n
≤20 mm	26	5	0.534
>20 mm	23	8
Pathological stage, n
I+II	45	9	0.052
III	4	4
Lymph node status, n
Positive	9	7	0.075
Negative	26	5
Not tested	14	1
Lymphatic invasion, n
Positive	41	12	0.670
Negative	8	1
Venous invasion, n
Positive	27	10	0.210
Negative	22	3
Nottingham histological grade, n
1+2	22	8	0.357
3	27	5
Ki-67 labeling index, n
High (≥40%)	28	9	0.506
Low (<40%)	18	3
Not tested	3	1
Stromal TILs, n
LPBC	16	3	0.737
Non-LPBC	33	10
Adjuvant chemotherapy, n
Performed	31	3	0.040
Not performed	17	8
Undetermined	1	2

[i] LPBC, lymphocyte predominant breast cancer; TILs, tumour-infiltrating lymphocytes.

Regarding TIM-3 expression, 30 patients (48.4%) were classified as TIM-3-positive (Fig. 2). Table II depicts associations between TIM-3 expression and clinicopathological factors in this cohort. Larger tumour size was associated with TIM-3-negativity (P<0.001), whereas LPBC correlated with TIM-3-positivity (P=0.013). However, there were no associations between TIM-3 expression and the clinicopathological factors, including age, menopausal status, administration of adjuvant chemotherapy, staging, Nottingham histological grade, lymph node status, lymphovascular invasion, or Ki-67 LI.

Figure 2. Immunohistochemical staining for T-cell immunoglobulin mucin-3. Positive staining is seen for lymphocytes (magnification, ×400).

Table II. Association between clinicopathological factors and TIM-3 expression.

Table II.

Association between clinicopathological factors and TIM-3 expression.

Factors	TIM-3-positive (n=30)	TIM-3-negative (n=32)	P-value
Median age ± SD, years	63±15	67±14	0.313
Menopausal status, n
Premenopausal	5	4	0.724
Postmenopausal	24	28
Unknown	1	0
Tumour size, n
≤20 mm	22	9	<0.001
>20 mm	8	23
Pathological stage, n
I+II	29	25	0.054
III	1	7
Lymph node status, n
Positive	6	8	0.752
Negative	17	16
Not tested	7	8
Lymphatic invasion, n
Positive	23	30	0.077
Negative	7	2
Venous invasion, n
Positive	15	22	0.195
Negative	15	10
Nottingham histological grade, n
1+2	15	15	>0.999
3	15	17
Ki-67 labeling index, n
High (≥40%)	16	9	0.062
Low (<40%)	12	21
Not tested	2	2
Stromal TILs, n
LPBC	14	5	0.013
Non-LPBC	16	27
Adjuvant chemotherapy, n
Performed	19	15	0.295
Not performed	10	15
Undetermined	1	2

[i] LPBC, lymphocyte predominant breast cancer; TILs, tumour-infiltrating lymphocytes; TIM-3, T-cell immunoglobulin mucin-3.

Correlations between galectin-9 or TIM-3 expression and relapse-free survival after surgery

RFS was superior for galectin-9-positive relative to -negative patients (60-vs.-44 months, P=0.027) as shown in Fig. 3A. For TIM-3-positive-vs.-negative patients, these values were 63 and 54 months (Fig. 3B, P=0.007).

Figure 3. RFS of patients with triple-negative breast cancer. (A) RFS of galectin-9-positive (green) or -negative (blue) patients. (B) RFS of TIM-3-positive (green) or -negative (blue) patients. (C) RFS of galectin-9(+)/TIM-3(+) (double positive) (orange), galectin-9(−)/TIM-3(+) (green), galectin-9(+)/TIM-3(−) (red) and galectin-9(−)/TIM-3(−) (double negative) (blue) patients. RFS, relapse-free survival; TIM-3, T-cell immunoglobulin mucin-3.

Impact of positivity for both galectin-9 and TIM-3 on clinicopathological features

Correlations between galectin-9- and TIM-3-positivity are shown in Table III, indicating a lack of association between galectin-9 and TIM-3 expression (P=0.548). Eight patients (12.9%) were both galectin-9- and TIM-3-negative (double negative), and 25 (40.3%) were positive for both (double-positive). The remaining 29 patients were either galectin-9- or TIM-3-single-positive.

Table III. Association between galectin-9 and TIM-3 expression.

Table III.

Association between galectin-9 and TIM-3 expression.

	TIM-3
Galectin-9	Positive, n	Negative, n
Positive	25	24
Negative	5	8

[i] P=0.548. TIM-3, T-cell immunoglobulin mucin-3.

Table IV summarizes correlations between the galectin-9/TIM-3 double-negative group and clinicopathological factors in the present cohort. Only larger tumour size and higher Ki-67 LI correlated with double-negative status (P=0.029 and 0.020, respectively) but there were no associations with age, menopausal status, presence of adjuvant chemotherapy, staging, Nottingham histological grade, lymph node status, lymphovascular invasion, or stromal TILs.

Table IV. Association between clinicopathological factors and galectin-9 and TIM-3 expression.

Table IV.

Association between clinicopathological factors and galectin-9 and TIM-3 expression.

Factors	Galectin-9 and TIM-3-negative (n=8)	Galectin-9 and/or TIM-3-positive (n=54)	P-value
Median age ± SD, years	72±10	64±15	0.166
Menopausal status, n
Premenopausal	0	9	0.590
Postmenopausal	8	44
Unknown	0	1
Tumour size, n
≤20 mm	1	30	0.029
>20 mm	7	24
Pathological stage, n
I+II	5	49	0.059
III	3	5
Lymph node status, n
Positive	3	11	0.670
Negative	5	28
Not tested	0	15
Lymphatic invasion, n
Positive	8	45	0.580
Negative	0	9
Venous invasion, n
Positive	7	30	0.128
Negative	1	24
Nottingham histological grade, n
1+2	4	26	>0.999
3	4	28
Ki-67 labeling index, n
High (≥40%)	7	30	0.020
Low (<40%)	1	20
Not tested	0	4
Stromal TILs, n
LPBC	1	18	0.416
Non-LPBC	7	36
Adjuvant chemotherapy, n
Performed	3	31	0.691
Not performed	3	22
Undetermined	2	1

[i] LPBC, lymphocyte predominant breast cancer; TILs, tumour-infiltrating lymphocytes; TIM-3, T-cell immunoglobulin mucin-3.

Correlation between galectin-9 and TIM-3 combined expression and relapse-free survival after surgery

Fig. 3C shows RFS for double-negative, double-positive or single-positive patients. The median RFS time of galectin-9/TIM-3-double-positive, galectin-9-positive/TIM-3-negative, galectin-9-negative/TIM-3-positive, and galectin-9/TIM3-double negative patients was 64, 57, 61, and 42 months, respectively. Thus, double-positive patients had a better prognosis, and double-negative patients had the worst prognosis (P=0.011).

Prognostic significance of galectin-9 and TIM-3 expression

According to univariate analysis, the presence of lymph node metastasis (P=0.004), galectin-9 negativity (P=0.039), TIM-3 negativity (P=0.029), and galectin-9/TIM-3 double-negativity (P=0.020) were each significantly correlated with poor RFS (Table V). Multivariate Cox proportional hazards analyses showed that galectin-9/TIM-3 double-negativity was an independent predictor of poor prognosis [hazard ratio (HR) 3.627, 95% confidence interval (CI) 1.037-12.68; P=0.044] (Table V). Additionally, lymph node metastasis was an independent risk factor for poor RFS (HR 5.925, 95% CI 1.555-22.58, P=0.009).

Table V. Univariate and multivariate analyses of relapse-free survival of patients with triple-negative breast cancer.

Table V.

Univariate and multivariate analyses of relapse-free survival of patients with triple-negative breast cancer.

	Univariate analysis			Multivariate analysis
Factor	HR	95% CI	P-value	HR	95% CI	P-value
Tumor size (>20 vs. ≤20 mm)	2.660	0.706-10.03	0.148
Lymph node status (positive vs. negative)	6.891	1.825-26.02	0.004	5.925	1.555-22.58	0.009
Nottingham histological grade (3 vs. 1+2)	1.829	0.535-6.256	0.336
Ki-67 labeling index (high vs. low)	1.497	0.387-5.793	0.559
Stromal TILs (LPBC vs. non-LPBC)	0.470	0.101-2.175	0.334
Adjuvant chemotherapy (performed vs. not performed)	0.358	0.104-1.225	0.102
Galectin-9 (negative vs. positive)	3.508	1.068-11.52	0.039	2.736	0.809-9.253	0.106
TIM-3 (negative vs. positive)	9.888	1.265-77.27	0.029	7.141	0.905-56.33	0.062
Galectin-9 and TIM-3 (double negative vs. others)	4.321	1.260-14.82	0.020	3.627	1.037-12.68	0.044

[i] 95% CI, 95% confidence interval; HR, hazard ratio; LPBC, lymphocyte predominant breast cancer; TILs, tumour-infiltrating lymphocytes; TIM-3, T-cell immunoglobulin mucin-3.

Discussion

Recently, the importance of TIM-3 in cancer immunology has been increasingly recognized due to its role as a checkpoint receptor inhibiting cytotoxic T cells (30–32). A previous meta-analysis implicated TIM-3 expression as an independent risk factor predicting poor overall survival (OS), but not cancer-specific and disease-free survival, in different malignant tumours (49). It was hypothesized that interactions of TIM-3 with its ligands, including galectin-9, results in the inhibition of both T cell responses and natural killer cell-mediated tumour cell cytotoxicity, resulting in the dampening of anti-tumour immunity and thence tumour escape (17). TIM-3 is believed to be expressed by exhausted T cells, the presence of which is associated with poor prognosis in several different cancers (49,50). In contrast, as mentioned above, TIM-3-positivity was associated with more favourable OS in patients with TNBC (38,49), although this conclusion is based only three studies which investigated whether TIM-3 expression predicted prognosis in TNBC. Table VI presents details of these previous studies, together with the current study (36–38). According to our results, TIM-3 expression on TILs from TNBC correlates with a more favourable prognosis. The reason for differences in the prognostic relevance of TIM-3 expression for TNBC as opposed to other cancer entities remains to be established. In this context, Burugu et al (37) suspected that it might reflect a more potent recognition of cancer cells by the immune system. The immune response of TILs expressing TIM-3 to tumour cells might be different in the tumour microenvironment of TNBC compared to that of other cancers. Therefore, additional studies examining the molecular mechanisms underlying the immune response of TILs expressing TIM-3 to carcinoma cells in TNBC are needed in order to explain this difference.

Table VI. Summary of the relationship between TIM-3 expression and prognosis of patients with triple-negative breast cancer.

Table VI.

Summary of the relationship between TIM-3 expression and prognosis of patients with triple-negative breast cancer.

First author/s, year	Patients, n	Prognosis	(Refs.)
Cabioglu et al, 2021	61	No prognostic significance was noted (using operative specimens after neoadjuvant chemotherapy), although TIM-3 expression was associated with a worse chemotherapy response	(36)
Burugu et al, 2018	387	TIM-3 expression was associated with good disease-free and overall survival	(37)
Byun et al, 2018	109	TIM-3 expression was associated with good cancer-specific survival	(38)
Present study	62	TIM-3 expression was associated with good relapse-free survival	-

[i] TIM-3, T-cell immunoglobulin mucin-3.

Additionally, some other associations between clinicopathological features of TIM-3 expression in TNBC patients have been reported by other investigators. One study included 30 TNBC patients, reporting that TIM-3 expression by TILs correlated significantly with the presence of lymph node metastasis and a higher Ki-67 LI, but its prognostic significance was not discussed (51). It may be important to note that TIM-3 expression is more frequent in TNBC than in other forms of breast cancer (51,52) and is also associated with higher PD-L1/PD-1 expression (37,38). Furthermore, consistent with the results presented here, it has also been noted that TIM-3-positivity is associated with the presence of abundant TILs (38).

Here, we have also demonstrated that negativity for the TIM-3 ligand galectin-9 is a poor prognostic factor, but this association lost significance in the multivariate analysis. However, we did find that TIM-3/galectin-9-double-negativity remains significantly predictive of poor prognosis in such a multivariate analysis. Galectin-9 on tumour cells is also a key protein that negatively regulates T cell function, leading to suppression of anti-cancer immune surveillance (21,26,27,53). Using breast cancer cell lines it was found that galectin-9 expression was associated with the suppression of anti-cancer immune surveillance (53). However, similar to our findings, some studies reported that galectin-9 expression predicts a favourable prognosis in breast cancer (34,54). Nonetheless, it must be recognized that there is a discrepancy between the generally reported immunosuppressive function of galectin-9 and its opposite prognostic significance in TNBC. Although galectin-9 ligation of TIM-3 pathway induces dysfunction of TILs [for example, in hepatocellular carcinoma (55)], here we found that it was the double-negativity for TIM-3 and galectin-9 that predicted a poor prognosis whereas positivity for both was associated with a more favourable course. The functional role of galectin-9 and TIM-3 in anti-tumour responses might be different in TNBC than in some other types of cancer. It is clear that the TIM-3/galectin-9 pathway can suppress cytotoxic T cells and NK cells and protect the tumour (16), but it is also known that the presence of galectin-9 on breast cancer cells increases the strength of cell-cell interactions. This could thus prevent metastasis or at least reduce the metastatic potential of the tumour (56). As such, the outcome might be more favourable when breast tumour cells express galectin-9. To resolve this issue, additional molecular studies are needed, especially for TIM-3/galectin-9 double-negativity. Better understanding of the oncoimmunology of TNBC will hopefully lead to improved prognosis.

Some limitations of the present study must be recognized, including a relatively small sample size. Thus, additional studies with a larger number of participants must be performed. Second, this study used tissue microarrays to evaluate immunohistochemical staining for TIM-3 and galectin-9. This might have led to a bias in evaluating the expression of these proteins, despite the fact that we selected the most morphologically representative regions from the patients. Third, TIM-3 can be glycosylated, and glycosylated TIM-3 has a weaker ability to bind galectin-9 (57). Moreover, galectin-9 has three isoforms (58). This study analysed TIM-3 and galectin-9 expression by immunohistochemical methods using monoclonal antibodies which may have different specificities. Thus, the galectin-9 antibody used in this study reacts with all three isoforms of galectin-9 (59), so positivity for galectin-9 may reflect different isoforms of galectin-9. Although this method is versatile, further studies using Western blotting conducted on fresh tumour samples could provide further information on the roles of glycosylated TIM-3 and/or the different galectin-9 isoforms. Fourth, chemotherapy and/or ICB might influence the expression of TIM-3 and/or galectin-9. Anthracycline and taxane upregulate galectin-9 expression in some types of cancer cells (60)-although there were no patients receiving neoadjuvant chemotherapy in the present study. Nonetheless, additional studies should determine whether these drugs influence the expression of TIM-3 and/or galectin-9 in TNBC.

In conclusion, this study documented that TIM-3 or galectin-9 positivity predicts a more favourable prognosis in TNBC patients, in particular when the TILs are TIM-3-positive and the tumour is galectin-9-positive. Generally, the TIM-3/galectin-9 pathway is thought to suppress anti-cancer immunosurveillance, but here we reveal a positive influence on TNBC prognosis. However, the molecular mechanisms underlying the difference between TNBC and other cancers in this respect remain unclear and further analyses are needed to resolve this issue. This could contribute to improved therapy for patients with TNBC.

Acknowledgements

Not applicable.

Funding

The present study was supported in part by AMED (grant no. JP21lm0203006), the Osaka Community Foundation 2020, and research grants D1 and D2 from Kansai Medical University.

Availability of data and materials

All data generated or analysed during this study are included in this published article.

Authors' contributions

KY and MI conceived and designed the study. KY and MI were involved in immunohistochemical analyses. KY, MI, HY, KT, MS and TS acquired and analysed data. KY and MI drafted the manuscript, tables and figures. KY and MI confirm the authenticity of all the raw data. All authors read and approved the final manuscript.

Ethics approval and consent to participate

The present study was conducted in accordance with the Declaration of Helsinki, and the study protocol was approved by the Institutional Review Board of the Kansai Medical University Hospital (protocol no. 2019041; Hirakata, Japan). All data are completely anonymized. The Institutional Review Board waived the requirement of informed consent due to the retrospective design of the study, with no risk of identity exposure for the patients. The present study did not include any minors.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References